A0345
Title: Long memory in the marginalized time series of a VAR revisited
Authors: Tomas del Barrio Castro - University of the Balearic Islands (Spain) [presenting]
Philipp Sibbertsen - University of Hannover (Germany)
Andreu Sanso - University of the Balearic Islands (Spain)
Abstract: The purpose is to demonstrate analytically and through Monte Carlo results that the long memory observed in the marginalized univariate time series of a VAR(1), as defined by prior studies, depends not only on the value assigned to the main diagonal of the Toeplitz matrix associated with the parameters of the VAR(1) but also on the number of time series in the VAR. Using the damped trend representation proposed by a recent study, it is shown that as the number of time series in the VAR(1) increases, the long memory in the marginalized univariate time series decreases. The analysis is also extended to VAR(2) models, allowing for long memory associated with harmonic frequencies in the marginalized time series. Finally, the marginalized time series in these VAR models are pointed out to be cointegrated with each other, as the long memory behavior is governed by the damped trend.
